Building Smarter Application
What is a smart application?
Rule-based
Riged
Hard coded
Model-based
Evolves
Learning process
Objective
Combine Application with models
Architecture
How it works
Installation
Current environment
○ Ubuntu 16.04
○ Java 8
○ R
Terminal commands
$ cd ~/Downloads
$ unzip h2o-3.10.4.6.zip
$ cd h2o-3.10.4.6
$ java -jar h2o.jar
Implementation
H2O offers two ways for data prediction:
1. Using Code such as R, Java, Python, Coffeescript/Javascript, and Scala
2. Using the web interface Flow.
Question
What answer we’re looking for?
Would this client be
eligible for a loan and how
much interest would one
pay?
Data
● Club loans dates from 2007 to 2015 in United States
● The data used in the implementation is clean
● 163,987 Frames (instance or row)
● 15 Features (column)
Data Description in Flow
Simple Code in R
Loan Customer Implementation
$ gradle wrapper
$ ./gradlew build
$ ./gradlew jettyRunWar -x generateModels
Visit http://localhost:8080

H2O